Predicting temperature in blast furnaces using machine learning regression methods
Ano de defesa: | 2023 |
---|---|
Autor(a) principal: | |
Orientador(a): | |
Banca de defesa: | |
Tipo de documento: | Dissertação |
Tipo de acesso: | Acesso aberto |
Idioma: | por |
Instituição de defesa: |
Universidade Federal do Espírito Santo
BR Mestrado em Informática Centro Tecnológico UFES Programa de Pós-Graduação em Informática |
Programa de Pós-Graduação: |
Não Informado pela instituição
|
Departamento: |
Não Informado pela instituição
|
País: |
Não Informado pela instituição
|
Palavras-chave em Português: | |
Link de acesso: | http://repositorio.ufes.br/handle/10/16772 |
Resumo: | In the iron and steel industry, the stable operation of blast furnaces with efficient hot metal temperature monitoring and control is a very important task in the process to generate high-quality hot metal. In general, the operation of blast furnaces mostly relies on experience based decisions of human operators, which use the most recent measures of hot metal temperature and other operational variables to execute control decisions. However, due to the large number of variables and complex interaction among them, the operation of such equipment is not an easy task. This work proposes a prediction system as the first step of a larger and more complex control system for improving the efficiency of iron production considering the scenario in Brazil. It compares several machine learning models (K-Nearest Neighbors, Linear Regression, Extreme Boosting Machine, Light Gradient Boosting Machine, Random Forest, Support Vector Machine, XGBoost, and Multilayer Perceptron) in the task of hot metal temperature prediction. A good temperature prediction system will allow to better plan the control actions ahead in order to stabilize the furnace temperature during hot metal production. The proposed method was evaluated using real-world data from an steel-producing company. Results shown that the system can predict the hot metal temperature with mean absolute error of 9.56 when compared to the baselines with mean average error of 12.61. |